def check_index(self):
"""Function to check bowtie index.
Returns True is index exist on disk.
"""
if hasattr(self, 'bowtie2_index'):
return (pu.check_bowtie2index(self.bowtie2_index))
return False
def __init__(self, bowtie2_index, **kwargs):
"""Bowtie2 constructor. Initialize bowtie2 index and other parameters.
"""
super().__init__()
self.programName = "bowtie2"
self.dep_list = [self.programName]
if not pe.check_dependencies(self.dep_list):
raise Exception("ERROR: " + self.programName + " not found.")
self.valid_args = [
'-x', '-1', '-2', '-U', '--interleaved', '-S', '-b', '-q',
'--tab5', '--tab6', '--qseq', '-f', '-r', '-F', '-c', '-s', '-u',
'-5', '-3', '--trim-to', '--phred33', '--phred64', '--int-quals',
'--very-fast', '--fast', '--sensitive', '--very-sensitive',
'--very-fast-local', '--fast-local', '--sensitive-local',
'--very-sensitive-local', '-N', '-L', '-i', '--n-ceil', '--dpad',
'--gbar', '--ignore-quals', '--nofw', '--norc', '--no-1mm-upfront',
'--end-to-end', '--local', '--ma', '--mp', '--np', '--rdg',
'--rfg', '--score-min', '-k', '-a', '-D', '-R', '-I', '-X', '--fr',
'--rf', '--ff', '--no-mixed', '--no-discordant', '--dovetail',
'--no-contain', '--no-overlap', '--align-paired-reads',
'--preserve-tags', '-t', '--un', '--al', '--un-conc', '--al-conc',
'--un-gz', '--quiet', '--met-file', '--met-stderr', '--met',
'--no-unal', '--no-head', '--no-sq', '--rg-id', '--rg',
'--omit-sec-seq', '--sam-no-qname-trunc', '--xeq',
'--soft-clipped-unmapped-tlen', '-p', '--threads', '--reorder',
'--mm', '--qc-filter', '--seed', '--non-deterministic',
'--version', '-h', '--help'
]
#initialize the passed arguments
self.passedArgumentDict = kwargs
#if index is passed, update the passed arguments
if len(bowtie2_index) > 0 and pu.check_bowtie2index(bowtie2_index):
print("Bowtie2 index is: " + bowtie2_index)
self.bowtie2_index = bowtie2_index
self.passedArgumentDict['-x'] = self.bowtie2_index
else:
print(
"No Bowtie2 index provided. Please build index now to generate an index..."
)
def build_index(self,
index_path,
index_name,
*args,
threads=None,
overwrite=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""Build a bowtie2 index with given parameters and saves the new index to self.bowtie2_index.
Parameters
----------
index_path: string
Path where the index will be created
index_name: string
A name for the index
args: tuple
Path to reference input files
threads: int
Num threads to use
overwrite: bool
Overwrite already existing index
verbose: bool
Print stdout and std error
quiet: bool
Print nothing
logs: bool
Log this command to pyrpipe logs
objectid: str
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Options to pass to bowtie2. This will override the existing options in self.passed_args_dict (only replace existing arguments and not replace all the arguments).
:return: Returns the status of bowtie-build
:rtype: bool
"""
#check input references
if len(args) < 1:
pu.print_boldred(
"No reference sequence provided to bowtie2-build. Exiting")
return False
if not pu.check_files_exist(*args):
pu.print_boldred(
"Please check input reference sequences provided to bowtie2-build. Exiting"
)
return False
bowtie2_build_args = [
'-f', '-c', '--large-index', '--debug', '--sanitized', '--verbose',
'-a', '--noauto', '-p', '--packed', '--bmax', '--bmaxdivn',
'--dcv', '--nodc', '-r', '--noref', '-3', '--justref', '-o',
'--offrate', '-t', '--ftabchars', '--threads', '--seed', '-q',
'--quiet', '-h', '--help', '--usage', '--version'
]
#create the out dir
if not pu.check_paths_exist(index_path):
if not pu.mkdir(index_path):
print(
"ERROR in building bowtie2 index. Failed to create index directory."
)
return False
if not overwrite:
#check if files exists
if pu.check_bowtie2index(os.path.join(index_path, index_name)):
print(
"bowtie2 index with same name already exists. Exiting...")
self.bowtie2_index = os.path.join(index_path, index_name)
return True
bowtie2Build_Cmd = ['bowtie2-build']
if not threads:
threads = self.threads
newopts = {"--threads": str(threads)}
mergedopts = {**newopts, **kwargs}
#add options
bowtie2Build_Cmd.extend(
pu.parse_unix_args(bowtie2_build_args, mergedopts))
#add input files
bowtie2Build_Cmd.append(str(",".join(args)))
#add dir/basenae
bowtie2Build_Cmd.append(os.path.join(index_path, index_name))
#print("Executing:"+str(" ".join(hisat2Build_Cmd)))
#start ececution
status = pe.execute_command(bowtie2Build_Cmd,
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid)
if not status:
pu.print_boldred("bowtie2-build failed")
return False
#check index files
if not pu.check_bowtie2index(os.path.join(index_path, index_name)):
pu.print_boldred("bowtie2-build failed")
return False
#set the index path
self.bowtie2_index = os.path.join(index_path, index_name)
#return status
return True
def build_index(self, index_path, genome, objectid="NA"):
"""Build a bowtie2 index with given parameters and saves the new index to self.index.
Parameters
----------
index_path: string
Path where the index will be created
genome: string
Path to the reference genome
objectid : string
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
:return: Returns the status of bowtie2-build
:rtype: bool
"""
#check input references
if not _force:
if pu.check_bowtie2index(index_path):
pu.print_green(
"bowtie index {} already exists.".format(index_path))
self.index = index_path
return True
#check input files
if not (pu.check_files_exist(genome)):
pu.print_boldred(
"Please provide a valid input fasta file to build bowtie2 index"
)
raise ValueError("Please check input to star build index")
return False
bowtie2_build_args = [
'-f', '-c', '--large-index', '--debug', '--sanitized', '--verbose',
'-a', '--noauto', '-p', '--packed', '--bmax', '--bmaxdivn',
'--dcv', '--nodc', '-r', '--noref', '-3', '--justref', '-o',
'--offrate', '-t', '--ftabchars', '--threads', '--seed', '-q',
'--quiet'
]
#create the out dir
indexdir = pu.get_file_directory(index_path)
if not pu.check_paths_exist(indexdir):
if not pu.mkdir(indexdir):
raise OSError(
"Error creating bowtie2 index. Failed to create index directory."
)
return False
args = (genome, index_path)
internal_kwargs = {"--threads": self._threads}
#read build parameters
yamlfile = os.path.join(_params_dir, 'bowtie2_index.yaml')
if pu.check_files_exist(yamlfile):
yaml_params = pl.YAML_loader(yamlfile)
yaml_kwargs = yaml_params.get_kwargs()
internal_kwargs = {**yaml_kwargs, **internal_kwargs}
#add positional args
internal_kwargs['--'] = args
bowtie2Build_Cmd = ['bowtie2-build']
#add options
bowtie2Build_Cmd.extend(
pu.parse_unix_args(bowtie2_build_args, internal_kwargs))
#start ececution
status = pe.execute_command(bowtie2Build_Cmd, objectid=objectid)
if not status:
pu.print_boldred("bowtie2-build failed")
return False
if status:
if pu.check_bowtie2index(index_path) and not _dryrun:
#update object's index
self.index = index_path
if self.check_index():
return True
else:
raise OSError("Error building bowtie2 index")
return True
